The timestamp issue of Whisper #27
Description
Hello, when I was using Whisper and pyannote to complete a speech-to-text task, during runtime, part of the output was:
"Whisper did not predict an ending timestamp, which can happen if audio is cut off in the middle of a word. Also make sure WhisperTimeStampLogitsProcessor was used during generation."
Also, when recognizing my speech, there are instances where the speech from the same speaker is divided into several segments. I would like to know, should I use pyannote's segmentation method? Or can I still use Whisper's timestamps? If I want to use Whisper's timestamps, do you have any good suggestions? Or do you have any good advice for the issues I'm encountering?